ARP table not working properly?

Hi,

I’m using a rather strange configuration on my RB750 with ROS v6.19:

ether1 has 10.12.123.123 255.255.0.0
ether2 has 10.12.123.124 and the same subnet as ether1.
and I’m using a routing protocol.
My problem is that the routing protocol needs some time to start working properly and something stops working.

In this case I try to connect to the routerboard (using winbox&webfig), but it seems that the answer-packets are leaving through the wrong interface. When i disable ether2 i have no problem to connect using ether1. When i disable ether1 I have no problem to connect using ether2. But when both interfaces are active it does not work.

I cannot understand this, because only one interface should have the ARP entry of my notebook, so why the answer-packets get stuck in nowhere?

Would you mind sharing the reason for this strange configuration?

As a first shot I’d say adding both interfaces (and IP addresses) to a bridge should instantly solve your problem.
-Chris

These are Ptmp links where default forward on wlan is disabled and communication is handled by the routing protocol. We do not want to waste ip addresses or subnets (because we would need them for every link in this case).
The same configuration on openwrt works without a problem.

While it may be a bug that this doesn’t work, I would add a bridge interface with a management IP on it and use that to communicate with the box itself…